John Elihu Hall.

The American journal of science online

. (page 53 of 61)
Online LibraryJohn Elihu HallThe American journal of science → online text (page 53 of 61)
Font size
QR-code for this ebook


tube with vapor. During some of the measurements the vapor
escaped freely between the glass tube and aluminium cover and
burned there.

The remaining details and relative dimensions of the sulphur

Fig. 3.



/K



m



Fio. 3. Apparatus for direct determination of the boiUng point of salphar
with the nitrogen thermometer. Scale : 1 to 6*5.

vapor bath will be clear from the diagram (fig. 3), which is
drawn to scale. The arrangement shown in the figure is the
one with the aluminium shield.

The sulphur used was distilled to free it from a black residue
which was found in both of two different preparations of C.P.
sulphur. This residue, which is partly if not wholly ferrous
sulphide, would probably have had no appreciable effect on the
boiling point. After the close of these experiments the
sulphur boiling apparatus was sent to the Bureau of Standards,
where a number of measurements were made of the tempera-



Digitized by VjOOQ IC



with JSotling Point of Sulphur. 625

ture of the vapor within the aluminium shield compared with
the corresponding temperature in one of the sulphur baths
which Waidner and Burgess* have standardized and described
in connection with their measurements with the platinum
resistance thermometer. This comparison has been described
by Messrs. Dickinson and Miieller.f The diflferences amounted
in maximum to 0*04°, which may afford some measure of the
certainty with which the temperature of the sulphur vapor is
reproduced in an apparatus differing considerably from the
conventional form ana dimensions.

At the close of the measurements the gas thermometer bulb
and manometer were disconnected. In order to make certain
that our calculation of the volume of the unheated space con-
tained no unknown constant error, that portion of tliis space
lying outside the furnace was determined directly, as follows :
the merciiry was brought up to the fixed point with the capil-
lary open to the air. The connection between the two arms
of the manometer was then closed, and a known volume of air
(about 0*1^^) was drawn in through the capillary by drawing
off a weighed amount of mercury from the short arm. The
open end of the capillary was then sealed, the manometer con-
nection reopened, and the mercury again brought up to the
fixed point. The pressure necessary to effect this is a measure
of the volume of the space in the capillary and connections.
The results were :

Direct determinations 0'159"

Mean 0-J62««

Calculated in 1909 O-ieS^*'

The difference is negligible.

Finally, the location on the brass scale of the gas thermo-
meter of the "fixed point" which is situated in the top of the
short arm of the instrument and which defines its "constant
volume," was redetermined and found to have become dis-
placed by O'lS""™ since the initial determination in 1909. An
appropriate correction was accordingly made to this constant
in computing the results.

These various checks and verifications complete the observa-
tions. The more important constants of the instrument and
the formula used in the calculations are reproduced below from
the previous paper.:^

* BuH. Bur. Stds., vi, 150-280. 1909.

+ J. Washington Acad., ii, 176-180, 1912.

t Pub. No. 157, p. 62.



Digitized by VjOOQ IC



626 Day and Socman — Nitrogen Thermometer Scale

Material of the bulb : pure platinum 80 per cent
pure rhodium 20 per cent
Volume of bulb: F,=205-82«
Volume of unheated space: v, =0*309^®

-^-=000160

Expansion coefficient of bulb material: 10* /3=8*79 + 0'00161 1
Gas: pure nitrogen.

Initial pressure of gas (see Table I): /?^=about 600"".
Pressure coefficient of gas (see Table II) : a = 0*0036679 to
0-0036681.

Formula used in the computations:



-^..[.^(-rrv)-"-]



in which p^ and p are the corrected pressures at 0* and t respec-
tively. »

6. Experimental Results with Nitrate Bath.

Table I contains the nieasurementa of the zero-point of the
instrument before and after each heating and serves to show
that its variations are wholly within the accidental errors of
reading of barometer and manometer. In the third column
p\ is the measured initial pressure of the gas in millimeters
of mercury at zero degrees. The application of the correction
for the "unheated space" gives the pressures {p^ of column 4,

The measurements included in Table II are introduced
merely as a rough check on the pressure coefficient (a) of the
gas. A much more accurate determination has been made of
this constant for nitrogen by Chappuis* and his value of it
(0*0036681, corresponding to a pressure of 502™*"*) was used in
the calculations which follow. These measurements merely
afford a rough confirmation of this value within the somewhat
larger limits of error of our instrument. Columns 3 and 4
contain the pressures at the boiling point of water corrected as
in Table I.

In Table III, columns 3 and 4 contain the pressure measure-
ments near the three fixed temperatures of reference, reduced
as before ; t (column 5) is the gas thermometer temperature in
the nitrate bath and e (column 6) the corresponding electromo-
tive force of the thermoelement in microvolts ; e (column 7) is
the electromotive force of the same elements in benzophenone,
zinc and antimony respectively and the final column contains
the corresponding gas thermometer temperatures.
*Trav. Mem. Bnr. Int., vol. xiii, 1907.



Digitized by VjOOQ IC



with Boiling Point of Stdphur.
Table I. Measurements of j>«.



527



Date



1911

17 Oct.

19 "
21 "
21 "



Serial No.



Po



First Ocu Filling.



486-03
486-05
486-06
48605



Second Oas Filling.



Po



485-98
486-00
486-01
486-00



26 "


9


601-91


601-86


27 "


10


601-91


601-86


31 "


12


601-87


601-82


31 «.


13


501-88


601-83


2 Nov.


15


601-94*


601-89*


6 "


17


501-89


501-84


6 «


18


601-91


501-86


11 "


22


501-84


601-79


15 "


24


501-81


501-76


18 «


26


501-87


601-82


20 "


29


501-81


501-76


6 Dec.


30


501-79


601-74


8 "


35


501-84


501-79


11 "


38


501-82


501-77


22 "


43


501-83


601-78


27 "


48


501-77


501-72


1912








2 Feb.


54 i


501-86 1


601-81


16 "


61 1


501-88


601-83


Table II. Measurements of p, oo.



Date


1

1 Serial
No.

' 58
59

1 60

1


P'


P

683-69
683-67
6 83 -6^1


Barom.

' 754-3

1 753-7

753-4

1


i

t I


a


1912
16 Feb.


683-51
683-49
683-46


99-790
99-768 1
99-757
Mean|


0-0036684
0-0036685
0-0036686
0-0036685



* High wind, both mercury columns varying.



Digitized by VjOOQ IC



528 Day and Sodman — Nitrogen Thermometer Scale
Tablb III. Gas Thermometer Measurements in Nitrate Bath.



Date



Serial!
No. I




Temp.-of
fixed
point







Benzophenone


, boiling point.


1911












18 Oct.


2


1023-36


1024-47


306-97


2379.8


20 "


6


1020-60


1021-69


305-37


2364-7


30 "


11


1053-07


1054-18


304-91


14944*


1 Nov.


14


1054-84

-


1055-97


305-91


15002*



2369-2
2369-2
14996
15005
Mean



305-82-.
305-86
306-82
305-96



305-87



ZinCf melting point.



18 Oct.
20 "
4 Nov.

17 "

18 "
18 "



3


1216-27


1218-05


418-78


3429-4


5


1215-85


1217-61


418-53


3427-1


16


1256-96


1258-76


419-32


3432-2


25


1256-54


1268-35


419-18


3432-9


27


1255-85


1257-69


418-80


3429-0


28


1254-96


1256-80


418-29


3423-8



3433-7


419-24


3433-7


419-24


3433-7


419-48t


3433-7


419-27


3433-7


419-31


3433-7


419-36


Mean


419-28



Antimony y melting point.



10 Nov.


19


1627-10


1630-58


629-66


5625-8


5527-7


629-84


10 "


20


1627-09


1630-57


629-66


5525-2


5527-7


629-90


10 "


21


1626-96


1630-44


629-58


5524-2


5527-7


629-92


14 «


23


162508


1628-59


628-54


5615-6


5627-7
Mean


629-73




629-85



* Copper- Gonstan tan thermoelement,
f Wide temperature variation on manometer,
from the mean.



This valne is omitted



6. Interpolation Formula.

For convenient interpolation with thermoelements in this
temperature region, the following equation is accurate within
0*3° as far as the melting point ot copper, but may not be used
beyond that temperature or below the boiling point of ben-
zophenone :

e= — 308 + 8-2294 ^ + '001649 <"

In terms of the standard element nsed in our recent publica-
tion upon this subject,* this equation gives the following
electromotive forces :

• Publication No. 157, Carnegie Institution of Washington, pp. 109 et seq.
For the measurement and interpolation of temperatures above 1082-6°, the
reader is referred to the same pages ; for temperatures below 305*9" to page
536, following (Adams and Johnston).



Digitized by VjOOQ IC



with Boiling Point of Svlphur.



529



Fixed point


Temperature

(Nitrogen
Thermometer)


Standard

thertno-

element

(in

microvolte)


E. M. F.

calculated

(in
microvolts)


Difference
Obs. - Cal.




Mv.

+ 1


-3

+

+ 1

— 1



Degr.


Benzophenone

(boiling point)
Cadmium

(melting point)
Zinc

(melting point)
Antimony

(melting point)
Silver

(melting point)
Gold

(melting point)
Copper

(melting point)


805-9°
820-8
419-8
629-8
960-0
1062-4
1082-6


2365
2502
8429
5530
9118
10295
10534


2364
2502
3432
5529
9112
10296
10534


+ 0.1
0-0
-0-8
+ 01
+ 0-1
-0-1
0-0



7. Experimental Mesults on Sulphur.

Table \Y contains the measurements made in the sulphur
vapor bath, p^ and j? represent the gas thermometer pressures
corrected as before, t the resulting temperature, followed by
the barometer reading reduced to sea level at lat. 45°, and the
boiling temperature reduced to 760™*" pressure.



Table IV. Direct Measarement of Boiling Point of Sulphur.





Serial










t at


Date.


No.


P'


P


t


Barom.


760"'"


1912.












31 Jan.


49


1298-77


1301-07


443-21


746-3


444*45





60


1298-88


1301-16


443-26


747-0


444-43


a


51


1298-84


1301-17


443-27


747-0


444-44


iFeb.


52


1299-14


1801-44


443-42


748-4


444-47


((


53


1299-14


1301-44


443-42


748-2


444-48


3 Feb.


55


1299-70


1302-00


443-73


752-0


444-45


<(


56


1299-50


1301-81


443-62


751-4


444-40


u


57


1299-55


1301-87


443-66


751-0
Mean


444-47




444-45



Digitized by VjOOQ IC



530 Day and Sosman — Nitrogen Thermometer Scale



o

00







<2



n



o
pq



o

g



H

n



o









s




s
p



6.2


^


o


CO


1— 1


lO


Therm

dynam

scale


o


00


a>


•o


»o


^


^


'^


'^


'^


^


'^


^


'^


"^


-*


"V


'"^


"^


^


i 9.


f-H


o


CO


.


to


«> 1 S


00


l>


00


■^


-^














1 ' "


'"^


^


^


'^


'**•


•^


■^


-^


'^


'^


s 1 -^^


-*


-^


'*


"*


"^



»o

CO



00



CO

CO



o



I 't



o

>

i




1 5


-^

-^








T-t


1


O


00


CO


to

CO


6


II

o

a,






-*


CO





^2






-3 ©
o






e8



CO



o



»o



zoo

00



^ CO "^



o

CO

*o



o



CO (M »0 C^

CM r-^ (>4 O

CO CO CO *o




C "oj






o



o
O



o

O



c
o



o



§4







•^ o

^ g o

» © ►

E- I
^§-

>• 08 o

MS
■it

.2.S



T3

C



o

,£3



00






i 'c^


<


oj's:




=3^




! y





1




' l-H



•73
S
c8

QQ ^

2ffl



o
o



o
o

lb

s



^ o



^- 5



00

o



1^



I— •




« ■» I le^—



Digitized by VjOOQ IC



with Boiling Point of Sulphur. 531

In Table V are brought together, all the gas thermometer
determinations of the boiling point of sulphur since 1890 with
the necessary information for an intelligent comparison of the
determinations. Column 5 contains the initial pressure of the
gas used, and column 6 the original value published by the
author with a reference to the place of publication. Two
of these determinations were subsequently corrected by the
authors themselves. These corrections (with the reference)
are given in column 7.

Inasmuch as these various determinations were made under
somewhat different gas conditions, the results are not directly
comparable without reduction to some common unit. The
four columns which follow contain such reductions for pur-
poses of convenient comparison. Our own observations are
directly comparable with the numbers contained in the column
" Const, vol. nitrogen scale, p^ = 500." The fairest compar-
ison is afforded by reduction to the thermodynamic scale, in
view of the fact that the different gases used by the various
observers depart in varying degree from the behavior of a per-
fect gas expanding at constant volume from the same initial
pressure. It should be noted tiiat the small differences between
the four columns are really arithmetical and not experimental,
since none of the original determinations can claim an accuracy
closer than 0-05°.

The Calleudar and Griffiths determination above, which is
oftfen quoted as direct, is in reality indirect. In his first inves-
tigation (Phil. Trans. 1887) Callendar showed that his parabolic
formula represented within 1° tlie variation of the resistance of
platinum with the temperature as determined by the constant
pressure air thermometer as far as 600°. In this later work
(Phil. Trans. 1891) he showed by a comparison of two resist-
ance thermometers with the air thermometer, using sulphur
merely as a constant temperature bath, that his original value
of S = 1-67 would still represent the results for these ther-
mometers within the limits of error. This value of h was then
used to calculate the sulphur boiling point determined with
several platinum thermometers in the usual (Meyer) form of
sulphur boiling tube.*

* The air thermometer observations of CaUendar and Griffiths in snlphor
vapor comprise fourteen readinj^s, made between 4:52 and 5:30 p. m. on
Friday, September 12, 1890, and between 8:25 and 9:07 the foUowing morn-
ing, as foUow8(PhU. Trans., 182 A, 1891, p. 139) :

Friday, Sept. 12^ 1890. Saturdayy Sept. 13, 1890.



Time


Temperature


Time


Temperature


4:52


444-52"


8:25


444-51**


5:02


444-90


8:28


444-51


5:06


444-68


8:38


444-52


5:18


444-77


8:42


444-53


5:21


444-79


8:48


444-54


5:26


444-94


8:50


444-52


5:30


444-98


9:07


444:61



Digitized by VjOOQ IC



632 Day and Sosman — JHfiirogen Thermometer Scale

8. summary.
The new ^s thermometer temperatures which this investi-
gation has given ns are brought together in Table VI, ex-
pressed (column 2) in terms of nitrogen expanding at con-

Conceming the salphnr used with th« air thennometer, the foUowing
information ia offered (p. 140) :

'* The snlphor naed on the Friday was poured out of the apparatna before
it solidified. It was found to be much discoloured, owing to the presence jof
▼arions impurities due to residues of oil, red lead, etc., used in fitting the
iron tubes together. Fresh sulphur was used on the Saturday, and this, on
examination, appeared to have suffered hardly any change.'*

Notwithstanding the fact that the conditions sun*ounding the obserrations
of Saturday appear the more fav^orable, these are rejected, with this state-
ment (p. 140) :

" The mean values of t deduced from the observations with the platinum
thermometers Mi and Ma, by assuming the value <¥=l-570 in formula (d), are:
On Friday, t - 444-78' C.
On Saturday, t - 444-84' C.

** The mean values of i deduced from the simultaneous observations with
the ail thermometer are :

On Friday, t = 444-80'' C.
On Saturday, t = 444-52' C.

**The value found on Friday is seen to agree perfectly with that de-
duced from the observations with the platinum thermometers. The value
of t deduced from Saturday's observations is 0*32'' too low ; but it is prob-
able that the value of mk, used in reducing the observations taken on that
day, is a little too great."

*'..«. We may, therefore, conclude that these experiments, so
far as they go, are a complete verification of the value of 6 found in 1887,
and show that the platinum wire has not altered appreciably in the
interval."

Of the apparatus used, the following details are taken from another part
of the paper (p. 120) :

** Two new thermometers. Mj and Mi, were therefore constructed out of
the remainder of the old spiral [1887], and were very carefully compared
with the air thermometer at a temperature very near the boiling-point of
sulphur. . . . The result agrees perfectly with that found in 1887, and
shows that the value of the ^-coefficient has not altered appreciably in the
interval.

** The apparatus used for this comparison [the sulphur bath used with the
air thermometer], although useful as a constant high-temperature bath, was
not very well suited for determining the actual temperature of the sulphur
vapour. Another series of experiments was therefore undertaken in a Meyer
tube, which proved to be more convenient for the purpose.''

On page 145 occurs the following statement of the final result (obtained in
the Meyer tube with resistance thermometers only) :

'* Assuming 6 = 1*570 for thermometers L, Mi and Mi, we find for the
corresponding air- temperature the value

t_^ 444^53" a

We believe this to be within 01° of the true temperature of the vapour of
sulphur boiling freely under a pressure of 760™"."

The value assumed for 6 is entirely dependent upon the temperature
determined with the air thermometer, and can not be more accurate than
this determination. From the air thermometer measurements above quoted,
O'S"' would appear to be a fairer measure of the uncertainty of the final sul-
phur point.



Digitized by VjOOQ IC



with Boiling Point of Sulphur.



533



stant volume from an initial pressure of 760"*™ and (column 3)
in terms of the thermodynamic scale. These values replace
the corresponding temperatures published in our papers to
which reference has been made.



Table VI.^SumcARiZED Table.



Point



Benzophenone (Kahlbanm)
boiling pt. at 760»»"

Cadmium, melting pt.
Zinc, melting pt.

Stilphnr, boiling pt. at

Antimony (Kahlbanm),

melting pt.
Aluminum, melting point



Temperature



Const, vol. I Thermo-
Po = l at. I dynamic



805-86

820-8
419-3

444*40

629-8

658-5



805-9

320-9
419-4

444-55

680-0

658-7



Notes



Transferred by platin-
rhodium and copper-con-
stantan thermoelements.

Interpolated.

Transferred by thermo-
elements.

Direct.

Transferred by thermo-
elements.
Interpolated.



Finally a comparative table is added showing in terms of the
same (thermodynamic) scale a comparison of our results with
those obtained by Holbom and Henning in the latest work
published from the Reichsanstalt.



Table VII. — Compabative Table.
Thermodynamic Scale.



Point



I Holbom and Hen- | Day and Sosman
I ningl911 1912



Benzophenone


305-9


305-9


Cadmium


320-9


320-9


Zinc


419-4


419-5


Sulphur


444-61
Holbom and Day


444-56


Antimony


630-6


630-0



Geophysical Laboratory,
Carnegie Institution of Washington,
March, 1912.

Am. Joub. Sci.— Foubth Sbbies, Vol. XXXIII, No. 198.— June, 1912*
85



Digitized by VjOOQ IC



534 Adams and Johnston — Standard Scale of Temperatures.



Aet. XLIV. — A Note on the Standard Scale of Tempera-
tures between W0° and 1100^; by L. H. Aidams and J.
Johnston.

A YEAR ago, at the time when the original measurements
recorded in this note were completed, there was an outstand-
ing uncertainty of about 1° in the temperature scale around
400°; at the present time, by reason of the concordant results
obtained in the best series of gas thermometer determinations
within this region — those recently published by Holborn and
Henning* and Day and Sosmanf — tnis uncertainty is not more
than 0'1°, and is probably less than this, at temperatures up to
500°. The definite establishment of the temperature scale ren-
ders the conclusions presented in this note to some extent
supererogatory ; nevertueless, it has been thought worth while
to present them, as, at the least, they serve to confirm those
expressed in the preceding paper.

In what follows, we present independent thermoelectric
measurements at the boiling points of naphthalene and benzo-
phenone and of the freezmg points of four metals — tin, bis-
muth, cadmium and lead ; the agreement of these results with
the best resistance thermometer measurements of the same
fixed points shows that the thermocouple is not inferior to the
resistance thermometer as an accurate temperature-measuring
device within the temperature range in question. Moreovc,
we propose to show that the most thorough and most extensive
series of resistance thermometer measurements — those of
Waidner and Burgess, made at the Bureau of Standards — are
also in remarkable agreement over the whole range of
temperature (up to 1100°) with the gas thermometer measure-
ments of Day and Sosman, when they are expressed in the
same scale. At the same time this comparison shows that, if
we consider all of the points,^ excepting sulphur, to be fixed
by the gas thermometer work, and on this basis set up an
interpolation formula and calculate therefrom the boiling

{)oint of sulphur, the resistance thermometer measurements
ead to a value (444-55°) identical with the gas thermometer
determinations.

Calibration of the Thermoelements.

In connection with another inve8tigation§ it became neces-
sary to calibrate carefully some copper-constantan thermo-

* Ann. Physik, xxxv, 861-74, 1911. f Preceding paper.

X Namely, the boiling points of naphthalene and benzophenone, and the
freezing points of tin, cadmium, zinc, antimony, silver and copper.
§ This Journal (4), ixxi, 501-17, 1911.



Digitized by VjOOQ IC



Adams and Johnston — Standard Scale of Temperatures. 535

couples, which were then employed in determining the freez-
ing points of the metals tin, bismnth, cadmium and lead.*
This calibration has been described in the paper just referred
to ; but the account there given must be amplified by the fol-
lowing additions and corrections, which are rendered necessary
by the slight changes in the temperature scale resulting from
the new and more accurate gas thermometer determinations at
temperatures up to 500"^.

It has been foundf that too much reliance can not be. placed
on the readings of copper-constantan thermoelements at the
zinc point (419*4°), for some diminution of electromotive force
sets in, not serious, but sufficient to preclude the most accurate
measurement. For this reason we have ceased to make use of
the zinc point as a calibration temperature; for the same
reason, we give the calibration curve only up to about 360°.

The calibration temperatures, expressed in the corrected
scale, together with the corresponding values of the electro-
motive force of the standard element, are given in Table I.

Table I.— Calibration Temperatures.



t


e in microvolts


Difference


Observed


Calculated*


In microvolts


In degrees











0-0


0-0


25-00


979


980-0


— 1-0


-•025


50-00


2012


2012-7


+ 0-7


+ •016


7500


3096


3095-8


+ 0-2


+ 005


100-00


4227


4226-1


+ 0-9


+ •019


217-95


10119


10119-3


—0-3


— •005


3061*


15007


15007-0


0-0


0^0



1 The benzophenone ased was from Merck, which melts (at 46*9°) 0*8°
lower, and boils 0*2*' higher, than that obtained from Eahlbanm (Waidner
and Burgess, Bull. Bur. Standards, vii, 6). The f. p. of our Merck benzo-
phenone was also 46'9*' ; consequently we have added 0*2** to the accepted
b. p. (805*9°) of Eahlbaum benzophenone.

» From the equation e=88*105e +004442 «* -000002856 e«, obtained by
the method of least squares, e as measured at the zinc point (419*4°) was
21755 microvolts ; the equation gives 21688, which does not differ by more
than what one might expect in view of the extrapolation through more than
110°.

To reproduce the above data, a quadratic equation is insuf-
ficient, except over a very short range, and so is the cubic

*Loc. cit., p. 508.

f Cf . Day and Sosman, preceding paper, p. 621.



Digitized by VjOOQ IC



Online LibraryJohn Elihu HallThe American journal of science → online text (page 53 of 61)